Qualcomm's IMOD: A Technology Laud (But Is The Strategy Flawed?)

Last week’s OLED writeup reminded me that I hadn’t yet crafted another long-planned post, on Qualcomm’s competitive Interferometric Modulator (IMOD) display technology. I first heard about IMOD (which Qualcomm trademarks with the Mirasol marketing monicker) via a breakfast meeting with the company at the 2007 SIGGRAPH show. The MEMS-based concept, reminiscent of the structure that creates the iridescent colors of a butterfly wing:

is quite cool…although as you’ll soon see, I’m somewhat skeptical of the prospects for Qualcomm’s implementation success.

The fundamental IMOD pixel (or, for that matter, sub-pixel) structure employs two parallel reflective planes, the upper one semi-transparent and merged with a thin film layer. The lower plane is flexible:

In the IMOD’s ‘open’ state, destructive interference between the spatially shifted versions of light waves reflecting from the two planes cancel each other out. However, a specific plane spacing-dependent frequency experiences constructive interference, leading to its perception in the visible light spectrum. After the application of a switching ‘write’ voltage, the gap between the two planes collapses, leading to destructive interference of all visible-spectrum light frequencies and therefore to a perceived black pixel or sub-pixel. A hysteresis effect enables the IMOD, once transitioned from ‘open’ to ‘closed’ or visa versa via the ‘write’ pulse (a negative voltage in the ‘closed’ to ‘open’ transition situation), to indefinitely remain in its new state via a much smaller (translation: lower power) ‘bias’ voltage. Translation: no LCD constant refresh requirement.

Creating color involves an array of subpixels; 14 IMOD elements for each of red, green and blue in this conceptual image:

with varying ‘open’ gaps to generate red, green and blue constructive interference in the open state. Selectively opening and closing various IMODs at any point in time (i.e. spatial dithering) generates shades of each color, and temporal dithering (i.e. switching the IMODs into various pattern combinations much faster than the human eye can perceive each switch) further expands the perceived ’shades of grey’ range. To wit, Qualcomm claims that IMOD switches 1000x faster than LCD.

The technology’s pretty awesome as a MEMS poster child, I suspect you’ll agree, but then again over my years with EDN I’ve encountered lots of awesome technologies that never made the leap to market success because they weren’t cost-effectively manufacturable in high volumes with adequate yields, or for other reasons. Let’s begin with my technical concerns. Note that IMOD fundamentally relies on reflected ambient light. Unlike LCD (and like self-illuminating OLED), this characteristic means that it (theoretically) doesn’t need a backlight, with requisite power consumption and thickness advantages as a result.

It also means that the display is clearly viewable in bright sunlight, and when wearing polarized sunglasses. However, it also means that a deficit of ambient illumination means a deficient display. Qualcomm’s documentation points out that a sensor-controlled frontlight can be added…I’d respond that I can think of very few applications for which a frontlight wouldn’t be necessary, and with its inclusion come LCD-reminiscent incremental power consumption (to a degree dependent on how often the IMOD-inclusive device is used in dim light and the dark) and thickness.

Also, until recently the technology was only available in a bichrome implementation. Notice that I didn’t say monochrome (i.e. shades-of-grey)…I said bichrome, i.e. one color (yellow (PDF), for example), plus black. Certainly, plenty of low-end applications exist for which such a display would be adequate (albeit perhaps not optimal). But although Qualcomm can as a result quickly ramp volumes, it’s hard for me to envision the company being able to do so in a per-unit profitable manner given the comparative maturity of competitive alternatives.

What about the color displays that Qualcomm’s been promising ever since it acquired IMOD developer Iridigm in late 2004? As of May, the specifications of the first member of the family (PDF) are now public…and they’re pretty underwhelming both in an absolute sense and relative to OLED and LCD alternatives:

• 0.9" diagonal size
• 128×64 pixel resolution
• 160 ppi resolution
• 0.159mm pixel pitch
• 7:1 (!!!) contrast ratio
• 15 fps frame rate
• -10 to 60°C operating temperature range, and perhaps most telling of all…
• No specified color depth

Look back at my past OLED coverage and you’ll hopefully see that the root of my skepticism is the seeming insurmountable challenge of any new display technology to compete against the mature LCD juggernaut…except in applications where LCD manufacturers consciously choose to no longer compete! That same cynicism, albeit even further magnified, also exists in my mind regarding IMOD. The LCD industry is flush with companies determined to drive the technology forward both in an absolute sense and in vigorous competition with each other. No matter how much Qualcomm claims that IMOD can be manufactured in LCD fabrication facilities with little to no equipment and production flow alterations, and no matter that the company announced back in May that it’d jointly opened the first IMOD-dedicated fabrication facility…IMOD will need to cultivate a much broader and deeper ecosystem before I’ll seriously entertain the idea that it’s capable of going toe-to-toe with LCD and OLED.

Which leads to my final concern…and it’s a biggie. Excuse me while I point out the elephant in the room…Qualcomm has a sordid reputation when it comes to technology licensing agreements. I’m not faulting the company for negotiating hard on initial licensing deals…I’m talking about how Qualcomm is widely reputed to insist on re-negotiating (renewing, whatever…) those agreements heavily slanted in its favor as technologies become more popular, in a way that makes continued competition against Qualcomm an unprofitable endeavor (CDMA, anyone?). In light of past history, I’d frankly think twice (or more) before signing an IMOD-related agreement with the company, especially if I was a SoC competitor. To wit, it’s hard for me not to believe that if IMOD becomes popular, Qualcomm will build optimized interfaces for it into the company’s own chipsets.

While I (somewhat) buy the company’s argument that IMOD (potentially) advances the mobile phone, mobile television and other platforms in which Qualcomm plays, from a product standpoint IMOD is quite orthogonal to the company’s other lines. As such, while I can see how the 2004 acquisition of Iridigm made sense from a ‘deep pockets to drive the technology forward’ standpoint, I’d proffer that IMOD’s full potential won’t be realized until and unless Qualcomm re-spins the technology into a separate company capable of setting its own course from a business philosophy standpoint. Agree or disagree, readers? I welcome your thoughts.

p.s….The November 2007 issue of Scientific American included a pretty good article on IMOD. It’s only available for purchase on the magazine’s website, however a Google search reveals other sources (PDF in both cases)…